par Amerio, Pietro
Président du jury Gevers, Wim
Promoteur Cleeremans, Axel
Publication Non publié, 2025-12-01
Thèse de doctorat
Résumé : Since the inception of vision science in the early 19th century, scholars have shown interest for the human ability to perceive very brief visual stimuli. Research on this topic has been tightly linked to the development of presentation devices allowing short exposures. The field recently gained momentum thanks to the development of an LCD-based tachistoscope, capable of presenting arbitrary images for durations as brief as a few microseconds. This thesis capitalizes on such advancements to explore two broad issues related to visual perception. The first pertains to the integration of visual information over time, the second concerns the possibility of perception without awareness. The first part of the thesis is based on a well replicated finding – the main predictor of performance in a perceptual task is the product of stimulus intensity and exposure duration, a quantity often termed stimulus energy. In other words, performance is constant provided that total energy is maintained and regardless of how it is temporally distributed. This principle of intensity-duration reciprocity highlights the brain’s ability to collect and integrate information over time. Because it has been confirmed in so many contexts, it became known as Bloch’s law, after the name of its discoverer Adolphe-Moïse Bloch. Prompted by recent reports of violations of this principle, we leveraged a modern tachistoscope to reassess the temporal summation properties of the visual system. We addressed this issue across two studies: one focused on simple visual stimuli, the other on complex realistic stimuli, such as human faces, objects and scenes. Confirming Bloch’s law, we show that detection and discrimination of simple stimuli depend on their total energy. The same is true when the perceived intensity of complex stimuli is to be judged. However, when we required participants to extract meaningful information from them, the theory failed – complex stimuli presented for longer durations required additional energy to be processed. In the thesis, we discuss the implications of these discrepancies for the study and our understanding of the perception of realistic stimuli. The second part of the thesis focuses on subliminal perception – the possibility that stimuli manipulated so that they remain subjectively invisible can nevertheless be processed and influence subsequent behavior. In consciousness research, the possibility of subliminal perception, and the methods one can use to achieve subjective invisibility, constitute major issues. In this light, the tachistoscope offers an entirely novel manner of achieving invisibility compared to traditional masking methods. We addressed the possibility of perception without awareness in two studies. The first one investigated the issue using both traditional masking methods and brief unmasked stimulation. Comparable unconscious perception effects were retrieved with the two techniques, thus opening the venue to the study of unmasked unconscious information. The second study focused on the way awareness reports are computed from sensory evidence. On top of testing for unconscious perception under conditions of divided attention, our goal was to compare several models of awareness report production, which we developed in the previous study. Crucially, both studies leveraged a bias-free measure of awareness, geared to probe observers’ awareness of the feature of the stimulus that is relevant for the perceptual task. This measure is very sensitive towards conscious knowledge and thus makes for a conservative test of unconscious perception – across the two studies, we did not find convincing evidence thereof. Overall, this thesis makes a strong case for the use of brief visual stimulation to study the properties of the visual system. Further, we discuss future venues for research on information integration and subliminal perception.

Documents en relation

DI-fusion